Effect of flumazenil on diaphragm electrical activation during weaning from mechanical ventilation after acute respiratory distress syndrome

British Journal of Anaesthesia 114 (2): 269 75 (215) Advance Access publication 21 November 214. doi:1.193/bja/aeu374 Effect of flumazenil on diaphragm electrical activation during weaning from mechanical ventilation after acute respiratory distress syndrome H. Rozé 1,2,3 *, A. Germain 1, V. Perrier 1, A. Dewitte 1,4, O. Joannes-Boyau 1, C. Fleureau 1 and A. Ouattara 1,2,3 1 CHU de Bordeaux, Service d Anesthésie-Réanimation II, Unité de Réanimation polyvalente de la Maison du Haut-Lévêque, Hôpital Haut Lévêque, Avenue Magellan, F-336 Pessac, France 2 University of Bordeaux, Adaptation cardiovasculaire à l ischémie, U134, F-336 Pessac, France 3 INSERM, Adaptation cardiovasculaire à l ischémie, U134, F-336 Pessac, France 4 University of Bordeaux, Bioingénierie tissulaire, U126, F-33 Bordeaux, France * Corresponding author. E-mail: hadrien.roze@chu-bordeaux.fr Editor s key points The mechanism by which midazolam impairs diaphragm function is unclear. In this exploratory study, flumazenil increased diaphragm electrical activity in patients previously sedated with midazolam. Increased diaphragmatic electrical activity correlated closely with increases in tidal volume. This suggests that midazolam has a direct effect on the diaphragm, but more data are needed. Background. Diaphragm electrical activation (EAdi) and the ratio of tidal volume to EAdi (V T /EAdi) may provide clinical information on neuroventilatory efficiency (NVE) in patients being weaned from mechanical ventilation. We tested the hypothesis that residual sedation could interfere with respiratory recovery, by assessing the effects of flumazenil on EAdi and V T /EAdi ratio. Methods. This observational study included 13 patients breathing with pressure-support ventilation (PSV) after a long period of controlled mechanical ventilation (i.e..4 days) plus midazolam-based sedation for acute respiratory distress syndrome. EAdi and respiratory patterns were compared before and after a bolus of flumazenil, which was given because neurological status needed to be evaluated. Results. Flumazenil induced a significant increase in EAdi [+71 (41 123)%, P¼.2] and V T [+17 (8 32)%, P¼.5], resulting in significantly decreased NVE [234 (15 43)%]. The increased V T was significantly correlated with the increased EAdi (r¼.7, P¼.9). Conclusions. During weaning from mechanical ventilation, the diaphragmatic contribution to the breathing process may be reduced by residual midazolam-induced ventilatory depression. The increased EAdi with reversal of residual sedation was associated with a proportional increase in V T. These findings should be considered by the attending physician when interpreting daily EAdi and V T changes during weaning from mechanical ventilation. Keywords: electrical diaphragm activation; mechanical ventilation; sedation; tidal volume Accepted for publication: 11 August 214 The use of neurally adjusted ventilatory assist (NAVA) has enabled measurement and monitoring of diaphragm electrical activity (EAdi) at the bedside during any mode of ventilation. 1 EAdi gives clinicians the opportunity to continuously monitor neural respiratory drive to the diaphragm. 2 This is presented as a waveform, in which the amplitude relates to changes in motor-unit firing rate and recruitment. 3 Simultaneous recording of tidal volume (V T ) determine the V T to EAdi ratio, which expresses the ability of the neural drive to generate inspiratory volume. Consequently, the V T to EAdi ratio may be considered equivalent to neuroventilatory efficiency (NVE). Assessment of NVE allows us to quantify the imbalance between increased neural drive, respiratory load, and diaphragm weakness. NVE has been investigated to set PEEP levels under NAVA, with the goal of increasing NVE by minimizing the EAdi required for V T generation. 4 NVE monitoring has also been proposed to help the clinician decide the optimal time for extubation. 5 7 Interestingly, a daily increase in EAdi 7 8 and V T 8 during the spontaneous breathing trial (SBT) has been described, but without any variation in NVE, 7 which suggests residual sedatives have an effect. The consequences of midazolam on diaphragmatic, intercostal, and abdominal muscle function have been previously reported through the DPga/DPpl index, which is the ratio of gastric-pressure changes to the ratio of pleural-pressure changes during inspiration. 9 The possible activation of accessory muscles renders it difficult to accurately determine the contribution of diaphragm contraction on changes to DPpl and V T. 1 Thus, we have conducted an observational study to test the hypothesis that residual sedation induced by & The Author 214. Published by Oxford University Press on behalf of the British Journal of Anaesthesia. All rights reserved. For Permissions, please email: journals.permissions@oup.com

BJA Rozé et al. midazolam, which is still present at the beginning of the weaning process, may interfere with EAdi, and that diaphragmatic function may be recovered when the effects of sedative drugs are reversed. Therefore, the effects of flumazenil on EAdi, V T and NVE were evaluated in patients being weaned from mechanical ventilation after prolonged sedation, ensured by midazolam, and in whom rapid assessment of neurological status was required. Methods Study population This observational study was approved by our institutional review board (Comité de Protection des Personnes Sud-Ouest et Outre Mer III, Bordeaux, France) and was conducted in our 22-bed intensive care unit. Informed consent from all patients or next of kin was obtained. Patients under PSV, who had been under controlled ventilation and sedated with midazolam for at least 4 consecutive days (with Ramsay scores of.3) and without a contraindication to flumazenil, were included in this study. All sedative drugs had been stopped for at least 24 h and the patients had no signs of respiratory distress under PSV. The attending physician planned an injection of flumazenil, to assess neurological status, while the patients still had positive midazolam measurements in their blood. An initial.5 mg bolus of flumazenil was given i.v. and could be repeated once, if necessary, until the patient was awoken. Ventilatory management Patients were ventilated with pressure-support ventilation (PSV), using a Servo-I ventilator that integrated the NAVA module (Maquet Critical Care, Solna, Sweden). Attending physician chose the assist level under PSV. EAdi was registered via a feeding nasogastric tube with a multiple array of electrodes placed at its distal end (EAdi catheter, Maquet Critical Care). Initial placement of the tube was done according to anatomical considerations. Then, correct positioning of the EAdi catheter was ensured using a specific function of the ventilator, entitled EAdi catheter positioning. 11 Each EAdi peak and all corresponding respiratory parameters, such as respiratory rates (RRs), V T, airway pressures, and flow rates, were recorded before and after flumazenil (maximal or minimal value within 5 min after the bolus was given). The NVE at EAdi peak before injecting flumazenil and at just after the injection were also calculated. Data were collected on a computer through the Servo-I using specific software: RCR (Maquet Critical Care). Statistical analyses Data are expressed as means (SD) for normally distributed continuous variables, and medians (inter-quartile range, IQR) for non-normally distributed continuous variables. The normal distribution of continuous variables was assessed using the Kolmogorov Smirnov test. A paired Student s t-test and Wilcoxon s matched-pair signed-rank tests were used to compare means and medians, respectively. Correlations between parameters were tested using Pearson s correlation test on normally distributed variables (with confidence intervals) and Spearman s correlation test for non-normally distributed variables. All P-values were two-tailed, and a P-value of,.5 was required to reject the null hypothesis. Statistical analyses were performed with NCSS software (Statistical Solutions Ltd, Cork, Ireland). Results Between February 211 and April 213, we prospectively included 13 consecutive patients who were weaned-off mechanical ventilation using PSV and who had received flumazenil. The patients characteristics are summarized in Table 1. Nine of these patients (69%) were men. The median duration of mechanical controlled ventilation was 7 days (5 11) and the median duration of PSV between until flumazenil was injected was 1 day (1 2). All patients had a Ramsay score of.3 and only one patient received.5 mg of flumazenil twice. The mean plasma levels of benzodiazepine were 659 (674) ng ml 21. Effects of flumazenil on EAdi, V T, and NVE Flumazenil provoked a significant increase in EAdi and V T, and a decrease in NVE (Table 2). The median differences were +17 (8 32)% for V T (P¼.5), +71 (41 123)% for EAdi (P¼.2), and 234 (15 43)% ml mv 21 for NVE (P¼.2) (Fig. 1). These increases of EAdi and V T with flumazenil were not correlated to the level of plasma benzodiazepine (r¼.13; P¼.65 and r¼2.1; P¼.77, respectively). Correlations between EAdi and V T There was no correlation between EAdi and V T before and after the flumazenil injection (r¼.5, P¼.85 and r¼2.3, P¼.3, respectively). However, as shown in Figure 1, changes in V T induced by flumazenil, in each patient, were correlated to changes in EAdi (r¼.7, P¼.9) (Fig. 2). Effects of flumazenil on RR and Ti A significant increase in RR was noted after flumazenil was given [+11 (22)%, P¼.3]. This increased RR was associated with an almost significant decrease in Ti [2.7 (.12)%, P¼.7]. Ti was correlated with RR before and after flumazenil (r¼2.84, P¼.3 and r¼2.91, P,.1, respectively), but not to V T before and after flumazenil (r¼.47, P¼.1 and r¼.54, P¼.5, respectively). Moreover, the variation in V T was not correlated with the variation in Ti (r¼2.2, P¼.5). Effects of flumazenil on haemodynamics Significant increases in heart rate and systolic arterial pressure were observed after injecting flumazenil [+7 (8)%, P¼.7 and +17 (1)%, P,.1, respectively]. Correlations between NVE pressure support and arterial blood gases NVE before flumazenil was not correlated with the level of pressure support (r¼2.13, P¼.65), Pa O2 /FIO 2 (r¼.1, P¼.73), 27

Effect of flumazenil on diaphragm electrical activation BJA Table 1 Characteristics of the patients included in the study (n¼13). Data are expressed as mean (SD) or median (inter-quartile range). VCV, volume-controlled ventilation; PSV, pressure-support ventilation; ARDS, acute respiratory distress syndrome; SAPS 2, Simplified Acute Physiology Score Patients Aetiology Age (yr) Height (cm) SAPS 2 Plasma midazolam concentrations (ng ml 21 ) VCV (days) PSV (days) Pressure support (cm H 2 O) 1 ARDS pneumoniae 78 175 47 546 5 1 8 4 151 7.38 4.3 2 ARDS pneumoniae 79 17 35 6 9 1 1 8 27 7.42 4.8 3 ARDS pancreatitis 68 163 64 98 5 1 12 5 24 7.47 3.9 4 ARDS pneumoniae 58 157 7 1 5 1 13 4 223 7.4 5.8 5 ARDS pneumoniae 25 18 42 77 15 1 16 5 15 7.46 5.4 6 ARDS pneumoniae 68 17 56 66 11 2 12 5 151 7.5 4.1 7 ARDS peritonitis 65 17 53 2238 7 1 16 8 25 7.38 5.2 8 ARDS pneumoniae 66 161 58 1775 5 1 14 8 132 7.37 4.2 9 ARDS peritonitis 8 181 52 291 6 1 15 4 174 7.41 4.2 1 ARDS peritonitis 69 158 72 36 7 2 12 5 27 7.38 4.4 11 ARDS pneumoniae 48 175 48 258 22 1 22 6 12 7.39 6.6 12 Pulmonary fibrosis 77 164 31 53 1 2 9 2 152 7.41 5.5 13 ARDS pancreatitis 61 16 47 24 6 2 1 4 255 7.37 4.7 64 (15) 168 (8) 52 (12) 659 (674) 7 (5 1) 1 (1 2) 13 (4) 5 (2) 195 (56) 7.41 (.4) 4.9 (.8) PEEP (cm H 2 O) Pa O2 / FIO2 ph Pa CO2 Pa CO2 (r¼2.17, P¼.55), ph (r¼.23, P¼.35), and RR (r¼2.51, P¼.7). After flumazenil, NVE was correlated with RR (r¼.58, P¼.4) but was still not correlated with the level of pressure support (P¼.65), Pa O2 /FIO 2 (P¼.73), Pa CO2 (P¼.55), and ph (P¼.35). Discussion In the present study, the major finding is that, after prolonged ventilation and sedation, reversal of sedation by flumazenil significantly increases EAdi and V T, whereas NVE decreases. Moreover, this significant increase in V T was significantly correlated with the increased EAdi: DV T (ml)¼3.937 DEAdi (mv)+27.43. Residual sedative drugs probably do persist while weaning from mechanical ventilation. Our patients had the same fixed pressure assist, whereas flumazenil reversed benzodiazepineinduced sedation and inhibition of respiratory drive. The consequence was increased V T, resulting in a well-tolerated increase in the patient s contribution to breathing and EAdi. Effects of midazolam and flumazenil on diaphragm and respiratory patterns Numerous studies have described midazolam-induced respiratory inhibition in healthy volunteers. The primary effect of midazolam in humans is to significantly decrease V 12 T. This reduction in V T of 4% is associated with an 1 12 increased RR and a reduction in minute ventilation. Wilkinson and colleagues 13 found flunitrazepam binding sites in rat diaphragms, which could be receptors for the direct neuromuscular effect of benzodiazepines. It has been found, in vitro, that benzodiazepines inhibit diaphragm contractility with 1% depression of twitch tension. 14 Moreover, Fujii and colleagues 15 found a significant reduction in transdiaphragmatic pressure in dogs that had been given benzodiazepines. Molliex and colleagues 1 investigated the effects of midazolam and flumazenil on intercostal, diaphragmatic, and abdominal muscle functions using the DPga/DPpl index, and intercostal and abdominal electromyography in healthy volunteers. These authors reported decreased diaphragm contraction and increased contribution by accessory muscles to the load of the higher upper airway resistances caused by midazolam in these non-intubated patients. Interestingly, in this study, the injection of flumazenil allowed V T to return to baseline values. It should be pointed out that these authors did not measure the diaphragm electromyography. By measuring EAdi, which records electrical activation of the crural diaphragm, our study strongly suggests that midazolam inhibits intrinsic diaphragmatic contraction. Indeed, flumazenil induced a significant increase in EAdi and V T. Interestingly, the increase in EAdi was significantly correlated with an increase in V T when airway pressure remained stable. Of course, the contribution of the accessory muscle was not assessed, but none of the patients had any signs of respiratory distress from accessory muscle activation under PSV before an injection of flumazenil. Moreover, the work needed to breathe did not differ after flumazenil was injected: the patients were intubated and upper airway resistances remained stable. 16 The contribution of the accessory muscle occurs essentially when the work of breathing increases significantly. 17 The decrease in V T with midazolam probably cannot be explained by alterations to the pulmonary mechanics, as no differences in functional residual capacity and dynamic pulmonary compliance have been found. 12 Therefore, in our 271

BJA Rozé et al. Table 2 Individual data of patients (n¼13) before flumazenil and after bolus of flumazenil. Data are expressed as mean (SD) or median (IQR). RR, respiratory rhythm; EAdi, diaphragm electrical activity; V T, tidal volume; CR, cardiac rhythm. *P,.5 before flumazenil vs after bolus of flumazenil. All comparisons were performed by using a paired t-test for means and a Wilcoxon test for medians Pts RR (cycles min 21 ) Ti flow (s) EAdi (mv) VT (ml) VT (ml kg 21 ) VT/EAdi (ml mv 21 ) CR (beats min 21 ) SAP (mm Hg) Before After Before After Before After Before After Before After Before After Before After Before After 1 15 26.95.8 1.8 4 78 95 11.1 13.5 433.3 237.5 115 13 11 14 2 17 13.98 1.4 2. 3.2 45 652 6.8 9.9 225. 23.8 72 8 127 139 3 22 22 1..9 2.2 1 42 49 7.6 8.9 19.9 49. 85 85 119 161 4 12 13 1.38 1.26 3.2 7.1 56 741 11.3 14.9 175. 14.4 1 11 1 11 5 31 3.61.57 4.9 26 31 555 4.4 7.9 63.3 21.3 12 13 133 14 6 29 35.64.6 8.4 9.2 5 52 7.6 7.9 59.5 56.5 89 95 148 155 7 27 29.68.69 7. 12 4 45 6.1 6.8 57.1 37.5 114 18 19 121 8 31 35.54.48 11. 16 471 481 8.8 9. 42.8 3.1 91 115 126 163 9 32 4.64.59 26. 4 34 44 4.5 5.8 13.1 11. 73 85 133 155 1 3 29.65.7 12.5 28 34 45 6.7 8.9 27.2 16.1 9 85 17 12 11 34 38.64.6 38. 52 54 56 7.1 7.2 13.3 9.7 85 9 13 145 12 28 33 1.5.66 46. 59 533 593 8.8 9.8 11.6 1.1 1 117 13 166 13 25 25.86.92 59. 11 6 69 1.5 12.1 1.2 6.3 79 83 13 15 26 (7) 28* (8).85 (.27).77 (.24) 8.4 (3.2 26.) 16.* (9.2 4.) 477 (127) 578* (147) 7.8 (2.2) 9.4* (2.7) 57.1 (13.2 175.) 29.* (9.2 4.) 93 (16) 11* (18) 123 (13) 144* (18) EAdi ( V) V TE (ml) NVE (ml V 1 ) 5 4 3 2 1 8 6 4 2 2 15 1 5 Before flumazenil After flumazenil * Fig 1 Comparison of expired tidal volume, diaphragm electrical activity, and NVE before and after an injection of flumazenil. Data are expressed as the mean (SD) of expired tidal volume (V TE ) or the median (IQR) of diaphragm electrical activity (EAdi) and NVE; *P,.5 before flumazenil vs after a bolus of flumazenil. All comparisons were performed using a paired Student s t-test. study, the increase in V T after flumazenil, for the same airway pressure, cannot be explained by improved pulmonary mechanics. Conflicting results have been found regarding the reversal effect of flumazenil. 18 22 The fact that these studies had different protocols, doses and timings of injections, and measurements could partly explain these variations. Additionally, it may be difficult to demonstrate the effects of flumazenil when residual sedation includes opioids. Opioids induce respiratory depression with a significant decrease in RR through m- and d-receptors. 23 24 Gross and colleagues 25 found that, in the presence of concomitant opioid-induced depression of ventilatory drive, flumazenil only reversed the benzodiazepine-induced component of ventilatory depression. * * 272

Effect of flumazenil on diaphragm electrical activation BJA DV T (%) 5 rho =.71, P=.9 4 3 2 1 2 4 6 8 1 DEAdi (%) Fig 2 Correlation between DEAdi and DV T (P¼.9) was DV T (ml)¼3.937 DEAdi (mv)+27.43. EAdi, electrical diaphragm activity (mv); V T, tidal volume (ml). In our study, flumazenil increased RR and decreased inspiratory time. Consequently, the increased V T under PSV cannot be explained by an increased inspiratory time. With the rapid reversal of midazolam-respiratory inhibition, V T increased significantly and three patients had, transitorily, a V T of.12 ml kg 21 predicted body weight. The rapid increase in EAdi and diaphragmatic muscle function with flumazenil was responsible for the increased transpulmonary pressure, as airway pressure was fixed. This highlights the risk of overdistension when patients awake under PSV. 26 It would have been interesting to transitorily reduce the level of pressure assist under flumazenil in order to keep V T constant and to quantify the reduction in driving pressure for a given increase in EAdi. These three patients had V T around 1 ml kg 21 predicted body weight, before flumazenil injection. It is possible that the assist level chosen by the attending physician was too high for them. This illustrates the dependency of V T on the assist level with PSV and the possible uncoupling between inspiratory effort and V T. This might have been different with proportional assist. 27 Neuroventilatory efficiency The consequence of injecting flumazenil in this population of patients was a decrease in NVE. The patients conditions did not worsen with this reduction in NVE. We only found a significant correlation between NVE and RR after injection of flumazenil. Arterial blood gases and levels of pressure support were not correlated to NVE before or after an injection of flumazenil. The range of NVE values, for our patients, was large. Some patients had very low or high NVE values under PSV; the level of assist was chosen by the attending physician, and none of the patients had any clinical signs of respiratory distress, otherwise PSV would have been stopped. EAdi can sometimes be affected by interindividual anatomical differences, such as an increased distance between the electrodes in the oesophagus and/or because the crural diaphragm lowers the amplitude of EAdi. 15 It istherefore possible that, in this situation, its variation is more clinically relevant than its value. NVE reflects volumegenerated determinants, that is, respiratory drive, diaphragm function, and respiratory load. Respiratory drive determinants may explain why patients had increased EAdi and V T with flumazenil. Respiratory drive was inhibited by the presence of residual sedative drugs, which were still present during the weaning process. In our study, patients were maintained in a sedated state under controlled mechanical ventilation for at least 4 days before initiating the weaning-off process, this was done to inhibit respiratory drive. In a recently published report, there was no improvement in NVE during SBTs, but there were progressive and significant increases in V T and EAdi between the first failed and then successful SBTs. 7 This suggests an increased degree of wakefulness. Elimination of sedation decreases ventilatory depression and increases EAdi, and the consequences on NVE depend on the ability of EAdi to generate V T when the same level of pressure assist is used. Long periods of sedation for severe respiratory failure may have complex effects on the control and efficiency of breathing, and can alter the coupling between respiratory drive and motoneuron activation, which will generate a specific V T. NVE involves a complex set of interactions between the respiratory centres that feed signals to a central control mechanism, which, in turn, provides output to the effector muscles. 17 For example, in exercising chronic obstructive pulmonary disease (COPD) patients, dynamic hyperinflation reduces the transdiaphragmatic pressure for a given EAdi. 8 It would also be interesting to study the effects of other factors, other than sedation, such as acidosis and thoracopulmonary compliance, on NVE with NAVA. It is important to consider the following points when considering the clinical relevance of our study. Patients were sedated with benzodiazepine: the effect induced by other sedative drugs on EAdi and NVE could be different. A combination of benzodiazepine and an opioid was used during controlled mechanical ventilation. The ventilator-depressant effects of opioids are known to potentiate those of benzodiazepines; 28 consequently, the interaction of opioids could have limited the effect of flumazenil reversal in ventilatory depression and, thus, affected EAdi, V T, and NVE. No opioid antagonists were given to patients. The study population was specific and consisted of intubated patients after prolonged controlled mechanical ventilation for acute respiratory distress syndrome. Results are probably different with COPD patients in terms of the effect of midazolam 29 and NVE, as these patients have reduced pressure-generating capacities for a given drive because of disadvantageous diaphragmatic configuration. 3 NVE is also probably more efficient under proportional-assist modes (NAVA), as assist is delivered in synchrony and in proportion to EAdi. Indeed, the limitation of V T observed with increasing NAVA levels in humans couldsuggest that the Hering Breuer reflex is operative and reduces output from the respiratory controller at the same V T level, irrespective of the NAVA level. 27 31 The same study with NAVA would be interesting, as flumazenil might have different effects on patients control of V T.Moreover, 273

BJA Rozé et al. with NAVA, synchrony is improved as the flow is synchronized to EAdi, thus making NVE easy to measure. 32 PSV asynchrony, especially regarding the trigger for expiratory flow, will not always allow correct measurement of NVE changes. Clinical interest This study highlights the effects of residual midazolam (for sedation) on EAdi, V T, and NVE: these effects need to be taken into account at the bedside when interpreting the evolution of a patient s respiratory parameters during weaning from mechanical ventilation. EAdi could be an interesting tool at the bedside to assess and limit the deleterious effects of midazolam on the functioning of the diaphragmatic muscle. Conclusion During weaning from mechanical ventilation after respiratory failure, the effects of receiving prolonged mechanical ventilation and sedation with midazolam may cause residual midazolam-induced ventilator depression, which may contribute to a decreased diaphragmatic contribution to the breathing process. The increase in EAdi induced by reversal of residual sedation (using flumazenil) was associated with a proportional increase in V T and a reduced NVE. Thus, residual sedation should be taken into account by the attending physician when interpreting daily EAdi and NVE changes at the bedside during weaning from mechanical ventilation. Authors contributions Study concept and design: H.R.; acquisition of data: H.R., A.G., V.P., O.J.B., A.D., and C.F.; analysis and interpretation of data: H.R. and A.O.; drafting of the manuscript: H.R. and A.O. Acknowledgements The authors would like to thank Erwan Floch, PharmD (Newmed Publishing Services) for reviewing the manuscript. We would also like to thank all the physiotherapists for their assistance, and the nursing staff in the intensive care unit for their valuable cooperation. All authors are members of the medical staff in the department and played a key role in the study. Declaration of interest H.R. received honoraria for public speaking from Maquet. Funding Support was provided solely from institutional or departmental sources, or both. References 1 Sinderby C, Navalesi P, Beck J, et al. Neural control of mechanical ventilation in respiratory failure. Nat Med 1999; 5: 1433 6 2 Sinderby CA, Beck JC, Lindström LH, Grassino AE. Enhancement of signal quality in esophageal recordings of diaphragm EMG. J Appl Physiol 1997; 82: 137 7 3 Sinderby C, Spahija J, Beck J, et al. Diaphragm activation during exercise in chronic obstructive pulmonary disease. Am J Respir Crit Care Med 21; 163: 1637 41 4 Passath C, Takala J, Tuchscherer D, Jakob SM, Sinderby C, Brander L. Physiologic response to changing positive end-expiratory pressure during neurally adjusted ventilatory assist in sedated, critically ill adults. Chest 21; 138: 578 87 5 Dres M, Schmidt M, Ferre A, Mayaux J, Similowski T, Demoule A. Diaphragm electromyographic activity as a predictor of weaning failure. Intensive Care Med 212; 38: 217 25 6 Liu L, Liu H, Yang Y, et al. Neuroventilatory efficiency and extubation readiness in critically ill patients. Crit Care 212; 16: R143 7 Rozé H, Repusseau B, Perrier V, et al. Neuro-ventilatory efficiency during weaning from mechanical ventilation using neurally adjusted ventilatory assist. Br J Anaesth 213; 111: 955 6 8 Rozé H, Lafrikh A, Perrier V, et al. Daily titration of neurally adjusted ventilatory assist using the diaphragm electrical activity. Intensive Care Med 211; 37: 187 94 9 Goldman MD, Grassino A, Mead J, Sears TA. Mechanics of the human diaphragm during voluntary contraction: dynamics. J Appl Physiol Respir Environ Exerc Physiol 1978; 44: 84 8 1 Molliex S, Dureuil B, Montravers P, Desmonts JM. Effects of midazolam on respiratory muscles in humans. Anesth Analg 1993; 77: 592 7 11 Barwing J, Ambold M, Linden N, Quintel M, Moerer O. Evaluation of the catheter positioning for neurally adjusted ventilatory assist. Intensive Care Med 29; 35: 189 14 12 Morel DR, Forster A, Bachmann M, Suter PM. Effect of intravenous midazolam on breathing pattern and chest wall mechanics in human. J Appl Physiol Respir Environ Exerc Physiol 1984; 57: 114 1 13 Wilkinson M, Grovestine D, Hamilton JT. Flunitrazepam binding sites in rat diaphragm. Receptors for direct neuromuscular effects of benzodiazepines? Can J Physiol Pharmacol 1982; 6: 13 5 14 Driessen JJ, Vree TB, Booij LH, van der Pol FM, Crul JF. Effect of some benzodiazepines on peripheral neuromuscular function in the rat in-vitro hemidiaphragm preparation. J Pharm Pharmacol 1984; 36: 244 7 15 Fujii Y, Hoshi T, Uemura A, Toyooka H. Dose response characteristics of midazolam for reducing diaphragmatic contractility. Anesth Analg 21; 92: 159 3 16 Montravers P, Dureuil B, Desmonts JM. Effects of i.v. midazolam on upper airway resistance. Br J Anaesth 1992; 68: 27 31 17 Yan S, Kayser B. Differential inspiratory muscle pressure contributions to breathing during dynamic hyperinflation. Am J Respir Crit Care Med 1997; 156: 497 53 18 Mora CT, Torjman M, White PF. Effects of diazepam and flumazenil on sedation and hypoxic ventilatory response. Anesth Analg 1989; 68: 473 8 19 GrossJB, WellerRS, ConardP. Flumazenilantagonismofmidazolaminduced ventilatory depression. Anesthesiology 1991; 75: 179 85 2 Alexander CM, Teller LE, Gross JB. Slow injection does not prevent midazolam-induced ventilatory depression. Anesth Analg 1992; 74: 26 4 21 Mora CT, Torjman M, White PF. Sedative and ventilatory effects of midazolam infusion: effect of flumazenil reversal. Can J Anaesth 1995; 42: 677 84 22 Shalansky SJ, Naumann TL, Englander FA. Effect of flumazenil on benzodiazepine-induced respiratory depression. Clin Pharm 1993; 12: 483 7 23 Morin-Surun MP, Boudinot E, Gacel G, Champagnat J, Roques BP, Denavit-Saubie M. Different effects of mu and delta opiate agonists on respiration. Eur J Pharmacol 1984; 98: 235 4 274

Effect of flumazenil on diaphragm electrical activation BJA 24 Shook JE, Watkins WD, Camporesi EM. Differential roles of opioid receptors in respiration, respiratory disease, and opiateinduced respiratory depression. Am Rev Respir Dis 199; 142: 895 99 25 Gross JB, Blouin RT, Zandsberg S, Conard PF, Häussler J. Effect of flumazenil on ventilatory drive during sedation with midazolam and alfentanil. Anesthesiology 1996; 85: 713 2 26 Richard JCM, Lyazidi A, Akoumianaki E, et al. Potentially harmful effects of inspiratory synchronization during pressure preset ventilation. Intensive Care Med 213; 39: 23 1 27 Brander L, Leong-Poi H, Beck J, et al. Titration and implementation of neurally adjusted ventilatory assist in critically ill patients. Chest 29; 135: 695 73 28 Gross JB, Zebrowski ME, Carel WD, Gardner S, Smith TC. Time course of ventilatory depression after thiopental and midazolam in normal subjects and in patients with chronic obstructive pulmonary disease. Anesthesiology 1983; 58: 54 4 29 Vozoris NT, Fischer HD, Wang X, et al. Benzodiazepine drug use and adverse respiratory outcomes among older adults with COPD. Eur Respir J 214; 44: 332 4 3 Aubier M, Murciano D, Fournier M, Milic-Emili J, Pariente R, Derenne JP. Central respiratory drive in acute respiratory failure of patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 198; 122: 191 9 31 Colombo D, Cammarota G, Bergamaschi V, De Lucia M, Corte FD, Navalesi P. Physiologic response to varying levels of pressuresupport and neurally adjusted ventilatory assist in patients with acute respiratory failure. Intensive Care Med 28; 34: 21 8 32 Piquilloud L, Vignaux L, Bialais E, et al. Neurally adjusted ventilatory assist improves patient-ventilator interaction. Intensive Care Med 211; 37: 263 71 Handling editor: J. P. Thompson 275